PO.MCB06.03 · 分子与细胞生物学
全基因组 DNA 甲基化谱分析揭示雌激素受体阳性和阴性乳腺癌中不同的表观遗传图景及新型生物标志物
Genome-wide DNA methylation profiling reveals distinct epigenetic landscapes and novel biomarkers in estrogen receptor-positive and -negative breast cancers
该海报暂无可下载的资料
AACR 官方页面
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
背景：乳腺癌是一种分子异质性疾病，部分由具有诊断和预后潜力的亚型特异性 DNA 甲基化模式所驱动。我们旨在构建一个全面的甲基化数据集，以鉴定新型差异甲基化区域，界定区分乳腺癌与正常组织的全局甲基化模式，并表征雌激素受体（ER）阳性与 ER 阴性肿瘤之间的差异。
方法：我们通过处理来自 GEO 和 GDC 的 HumanMethylation450 和 EPIC 芯片平台（GEO 编号 GPL13534、GPL21145）的原始数据，构建了一个大规模甲基化数据集。该队列包括来自未经治疗的乳腺癌患者和健康病例的实体组织样本。数据预处理、BMIQ 标准化和质量控制使用 R 中的 minfi 和 wateRmelon 完成。探针被注释到功能性基因区域（TSS1500、TSS200、5'UTR、第一外显子、基因体和 3'UTR）。正常组、ER+ 组和 ER- 组之间的差异甲基化分析（Δbeta > 0.2）使用 Kruskal-Wallis 和 Mann-Whitney 检验并进行 Bonferroni 校正。使用受试者工作特征（ROC）分析并采用 5 折交叉验证（cvAUC）评估生物标志物潜力。使用 clusterProfiler 进行基因本体（GO）富集分析。
结果：基于 HM450K 和 EPIC 芯片，肿瘤相较正常组织表现出显著的、位置特异性的甲基化改变，并具有不同的 ER 阳性和 ER 阴性模式。ER 阳性肿瘤表现出普遍的启动子高甲基化（455 个高甲基化和 43 个低甲基化基因区域；Δbeta>0.2，p<0.05），富集于模式特化（HM450K：4.7 倍富集（FE），p<0.001；EPIC：9.3 FE，p<0.05）和食欲调节（HM450K：17.2，p<0.001；EPIC：34.3 FE，p<0.001）。ER 阴性肿瘤（两个平台共有的 354 个高甲基化和 23 个低甲基化基因区域，Δbeta>0.2，p<0.05）富集于同亲性细胞黏附（EPIC：16.2 FE，p<0.001）和嗅觉受体基因（EPIC：FE 12.2，p>0.001）。两个平台一致地鉴定出高置信度的差异甲基化区域，例如 ER 阳性中 NEURL2 基因体的高甲基化（cvAUC：HM450K 0.99，EPIC 0.99；Δbeta HM450K：0.43，EPIC：0.3）以及 ER 阴性肿瘤中 NKAPL 5'UTR 的高甲基化（cvAUC：HM450K 0.99，EPIC 0.99；Δbeta HM450K：0.47，EPIC：0.44）。
结论：本研究绘制了 ER+ 和 ER- 乳腺癌的甲基化组图谱，鉴定出表观遗传失调的亚型特异性模式。该数据集和交互式可视化工具可在 epigenplot.com 获取。
查看英文原文 English abstract
BACKGROUND: Breast cancer is a molecularly heterogeneous disease driven partly by subtype-specific DNA methylation patterns with diagnostic and prognostic potential. We aimed to construct a comprehensive methylation dataset to identify novel differentially methylated regions, define the global methylation patterns distinguishing breast carcinomas from normal tissue, and characterize differences between estrogen receptor (ER)-positive and ER-negative tumors.
METHODS: We curated a large-scale methylation dataset by processing raw data from HumanMethylation450 and EPIC array platforms (GEO accessions GPL13534, GPL21145) from GEO and GDC. The cohort included solid tissue samples from untreated breast cancer patients and healthy cases. Data preprocessing, BMIQ normalization, and quality control were performed using minfi and wateRmelon in R. Probes were annotated into functional gene regions (TSS1500, TSS200, 5'UTR, first exon, gene body and 3'UTR). Differential methylation analysis (∆beta > 0.2) between normal, ER+, and ER- groups was conducted using Kruskal-Wallis and Mann-Whitney tests with Bonferroni correction. Biomarker potential was evaluated using receiver operating characteristic (ROC) analysis with 5-fold cross-validation (cvAUC). Gene Ontology (GO) enrichment was performed with clusterProfiler.
RESULTS: Based on HM450K and EPIC arrays, tumors exhibited significant, location-specific methylation shifts versus normal tissue, with distinct ER-positive and ER-negative patterns. ER-positive tumors showed pervasive promoter hypermethylation (455 hyper-, and 43 hypomethylated gene regions; Δbeta>0.2, p<0.05), enriched for pattern specification (HM450K: 4.7 fold enrichment (FE), p<0.001; EPIC: 9.3 FE, p<0.05) and appetite regulation (HM450K: 17.2, p<0.001; EPIC: 34.3 FE, p<0.001). ER-negative tumors (354 hyper-, and 23 hypomethylated gene regions shared by platforms with Δbeta>0.2, p<0.05), and were enriched for homophilic cell adhesion (EPIC: 16.2 FE, p<0.001) and olfactory receptor genes (EPIC: FE 12.2, p>0.001). Both platforms consistently identified high-confidence, differentially methylated regions, such as hypermethylation in the NEURL2 gene body in ER-positive (cvAUC: 0.99 HM450K, 0.99 EPIC; Δbeta HM450K: 0.43, EPIC: 0.3) and the NKAPL 5'UTR in ER-negative tumors (cvAUC: 0.99 HM450K, 0.99 EPIC; Δbeta HM450K: 0.47, EPIC: 0.44).
CONCLUSIONS: This study mapped the methylomes of ER+ and ER- breast cancers, identifying subtype-specific patterns of epigenetic dysregulation. The dataset and interactive visualization tools are available at epigenplot.com.
利益披露 Disclosure
D. Müller, None..
B. Gyorffy, None.